Overhaul RMG-RMS Python-Julia dependencies

[hwpang]

No description provided.

[hwpang]

@mjohnson541 pointed out that this will likely break RMG's CI. Could you point out where in RMG CI do we need to change to avoid breaking it? cc @JacksonBurns

[hwpang]

@mjohnson541 Recording our offline discussion here: We should replace pyjulia with JuliaCall in RMG (see here: https://juliapy.github.io/PythonCall.jl/stable/pycall/). It should make the installation process easier. cc @JacksonBurns

[codecov]

Codecov Report

Attention: Patch coverage is 51.59574% with 91 lines in your changes missing coverage. Please review.

Project coverage is 48.42%. Comparing base (1f37bfd) to head (6ce8bfa).

❗ Current head 6ce8bfa differs from pull request most recent head 827a1a4

Please upload reports for the commit 827a1a4 to get more accurate results.

Files	Patch %	Lines
src/PhaseState.jl	61.76%	52 Missing ⚠️
src/ReactionMechanismSimulator.jl	0.00%	27 Missing ⚠️
src/Reactor.jl	44.44%	10 Missing ⚠️
src/Plotting.jl	0.00%	1 Missing ⚠️
src/fluxdiagrams.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #256      +/-   ##
==========================================
- Coverage   48.71%   48.42%   -0.30%     
==========================================
  Files          31       31              
  Lines        8313     8351      +38     
==========================================
- Hits         4050     4044       -6     
- Misses       4263     4307      +44     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[JacksonBurns]

@hwpang you can open a pull request against RMG-Py and change this line of the CI file (https://github.com/ReactionMechanismGenerator/RMG-Py/blob/eed950a389738e70894c25e1d8388bb20ef75947/.github/workflows/CI.yml#L155) to install this branch of RMS from your repository. We can then just see the failures live.

[hwpang]

RMS-RMG twin PR: ReactionMechanismGenerator/RMG-Py#2640

[mjohnson541]

I've done some verification on the juliacall end:

1. The primary trick I used to make pyrms from pyjulia works with juliacall
2. I think we can get around the need for diffeqpy, at least I think I circumvented our original reason for needing it...which should be nice because pyrms will no longer need the full DifferentialEquations.jl installed.

I think in terms of merge order this would look like: merge the RMS PR, merge a pyrms PR, build new binaries for pyrms, and then merge RMG PR.

[hwpang]

@mjohnson541 Can you make the pyrms PR since you've looked into this? Thanks

[mjohnson541]

@mjohnson541 Can you make the pyrms PR since you've looked into this? Thanks

Yes, I'm waiting until this PR is building properly, right now I wouldn't be able to test the pyrms PR properly.

[mjohnson541]

Thanks! This looks great! Just a few questions.

[hwpang]

@mjohnson541 I addressed your comment and had a question. I will write pretty commit messages after we agree on the changes

[hwpang]

Removing the duplicate export would cause the test to fail. I undo the change.

[hwpang]

@mjohnson541 This is ready for another review / approve. Need to clean up the commits after approval and coordinate to merge at the same time with ReactionMechanismGenerator/RMG-Py#2640.

[mjohnson541]

LGTM!